ave CLAUDE.md

Repository instructions for aveproject/ave, a project that defines behavioral categories for AI-agent components. They explain the record format and how to add a new record to the project.

In plain words
What is it for?
Use them before changing the repository, especially when adding or editing an AVE record or working with its schema and scoring rules.
Why use it?
They prevent contributors from confusing a defined behavioral category with a runtime scanner finding. They also provide the project’s required process and schema details.

Instructions file

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add instructions/aveproject/ave/claude-md
Clone the repo
git clone --depth 1 https://github.com/aveproject/ave
Per session 1,671 This file is loaded in full into every session.
When invoked 1,671 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.01671 $0.01671
Opus 5 $0.00835 $0.00835
Sonnet 5 $0.00334 $0.00334
Haiku 4.5 $0.00167 $0.00167

Measured 2d ago against content hash 0e66d3e227d7, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

ave CLAUDE.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

CLAUDE.md · 201 lines

How it starts

The opening of the file, as written. The whole thing — 201 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md — aveproject/ave

Read this file completely before touching anything. Single source of truth for how work happens in this repo.


Project

aveproject/ave — the behavioral classification standard for agentic AI components. An independent standard that bawbel-scanner implements. NOT a feature of the scanner.

  • Records: 56 published (schema_version 1.1.0)
  • Schema: schema/ave-record-1.1.0.schema.json
  • Scoring: OWASP AIVSS v0.8
  • Registry: aveproject.org
  • Public API: api.piranha.bawbel.io
  • Scanner: github.com/bawbel/scanner (reference implementation)

This repo contains DEFINITIONS, not detections. An AVE record defines a behavioral class. A scanner Finding is one detection that references an AVE record by ave_id.


The critical distinction — read this twice

AVE Record (this repo)              Finding (scanner repo)
──────────────────────              ──────────────────────
static behavioral class definition  runtime detection instance
authored once by a human            produced by every scan
one per behavioral class            one per detection
NO confidence field                 HAS confidence field
NO evidence_stage field             HAS evidence_stage field
declares confidence_baseline        assigns actual confidence
declares evidence_kind_default      assigns actual evidence_kind
declares detection_stage            reaches an actual evidence_stage

confidence is PER-DETECTION. It never lives in an AVE record. The record declares the BASELINE; the scanner assigns the ACTUAL value.


Record schema v1.1.0

Every record validates against schema/ave-record-1.1.0.schema.json.

15 required fields (once status is active or deprecated): ave_id · schema_version · status · published title · description · attack_class · severity · behavioral_fingerprint aivss · owasp_mcp indicators_of_compromise · remediation references · researcher

Draft submit-required core (status: "draft" only needs these 8): ave_id · schema_version · status · title · description · attack_class · behavioral_fingerprint · references

Read the full file on GitHub · 201 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 201 lines · 1,671 tokens per session scan A 0e66d3e227d7

Subscribe to this mod's changes

ave CLAUDE.md is an instructions file published in the GitHub repository aveproject/ave (17 stars, last pushed 3d ago), licensed Apache-2.0. It adds 1,671 tokens to every session, about $0.0084 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.